This post is more than 5 years old
46 Posts
0
2259
cx-300i - win2k8 server - offline disks appearing
Hi,
I am mapping LUNs from an EMC 300cx-i to a server running Windows 2008 server OS.
I create the LUNs
Create a storage group
Add LUNs to storage group
on server side
Configure network connection to SAN
install NaviAgent 6.24.2.5.0
Install Win2k8 Powerpath 5.5.sp1.b512 - unlicensed
Verify the host registers on the SAN
Add the host to the storage group
add 1 discovery portal
log in 1 target (A0) - specifying automatically restore and multi-path
add the drives on the windows side
At this point, everything is fine - the correct LUNs show up in storage manager, with no extras.
Then I add the remaining discovery portals for A1, B0, and B1 (in this case since the LUNs default to SPA
wait a few mins - Rescan disks - everything is still fine.
Then I start logging on the other targets. At this point, each time I log on a target (A1, B0, B1) with the same settings used for A0, another offline drive appears (so by the end, I have 3 by the end)
I dont know what these are or why they're there - they dont exist on my Win2k3 machines. Their presence seems to sometimes mess up
the identification of SNMP drive mibs.
Is my mapping procedure wrong somewhere? can anyone provide insight?
Thanks for your help!
christopher_ime
2K Posts
0
November 8th, 2013 00:00
I suspect that the order that you presented the additional paths will explain why you are experiencing this. You likely performed the following:
1) You registered the host with just the single path: A0
2) Added registered host and LUNs to the storage group
3) *Then* you logged in and registered the additional paths: A1, B0, B1
If my assumption is correct (even though the order of your steps suggest otherwise), what you need to understand is that once assigned to a storage group, (right or wrong) Navisphere doesn't automatically associate the new paths with any LUNs already presented to the host via a storage group assignment.
In this case, the offline devices you are seeing in diskmgmt.msc would be a 0 byte (virtual) LUNZ that is masked out only once you present a LUN (specifically assigned Host ID 0) via these paths. However, those paths need to be associated with the storage group as follows.
To update it, there are three options:
a) Remove host from storage group and readd
Of course, the host temporarily loses access to the LUN and that outage is as fast as you can remove/add the host
NOTE: The above is not a popular choice as there are non-disruptive alternatives so...
b) Select "Reconnect" from "Connectivity Status"
- Expand the appropriate "Storage Group" and then "Hosts"
- Right click on the registered host
- Select "Connectivity Status"
- Review the list of initiator records
We do advertise this as non-disruptive from the CLARiiON's perspective, but to absolutely verify that you only touch the new paths...
c) From "Engineering Mode" (if you are unfamiliar with the keystroke combinations and Engineering "password" a quick Google search will provide the steps)
- right-click on the Storage Group and select "Connect Hosts"
- From the "Hosts" tab you will see an "Advanced" button
- Click the "Advanced button, and you will see for each host there are checkboxes next to each registered path; simply check any non-associated paths to update the host registration
Then, as you are already familiar with, perform a rescan in Disk Management and verify that PowerPath is claiming the paths/devices:
powermt display dev=all
CLARiiON Support Forum
Vipin VK
812 Posts
0
November 7th, 2013 17:00
First of all consider moving this discussion to Clariion Support Forum ( Actions > Move ).
Verify Powerpath service is running and see it is able to see the devices from Clariion or not.
wolf8963
46 Posts
0
November 8th, 2013 06:00
First, thank you both for replying.
Chris,
My normal procedure (in the past) was to connect all portals and targets first, then add the host to the storage group. But since Id been having the strange results in server manager on the last couple, I did this one a little differently to see if it cleared things up.
The server I'm working on is in the build stage so I wouldn't have minded using method A, but opted for B instead, since I have a couple other servers with this weirdness and wanted to test the non-disruptive method.
I first captured the output of the 'powermt display dev=all' and pasted it in a file as a compare point.
Then I checked 'connectivity status' for that host on navisphere.
It indicated 3 paths under ~management, and 1 path under the server name.
I then hit reconnect. At this point a message appeared indicating that it would attempt to connect ALL available paths, appeared. So I am not sure what you mean by being sure to touch only the new paths - there are no options to select which paths to reconnect.
Since this was a server being built, I wasn't shy about proceeding. After reconnecting, all paths now show under the server name.
I went back to the server and reran 'powermt display dev=all' and captured it and compared it to the original file. The 2 files were identical.
update: I rechecked display dev=all later and it then showed the updates and included all paths
Then I opened storage management and did a refresh. I was concerned at this point because one of the off-line, 0-size drives now appeared with a size equal to one of the SAN LUNs. However I rescanned disks, then all anomaly disks disappeared and storage management appeared normal.
Further more, I had some robocopies running on the server at the time, and these were not disrupted by the process!
Thank you so much for the good information and help!
Also, I will attempt to move this thread to clariion.
wolf8963
46 Posts
0
November 8th, 2013 10:00
still battling drives showing all zeros in HRstorage for 2008.
I uninstalled Windows MPIO. Now I see a bunch of offline drives.
Do I need this or does powerpath supply its own MPIO (or does 2008 use Windows)?
(based on all the offline drives, I presume Powerpath uses Windows MPIO)
PS display dev=all finds no drives
kelleg
4.5K Posts
0
November 8th, 2013 12:00
With Windows 2008, the process that I follow is to install PowerPath first, then EMC Host Agent, then using the iSCSI initiator on the host, connect the iSCSI NIC to the array. Be aware that unlicensed PP only allows for one iSCSI NIC and two paths. On the array you should have SPA and SPB port in the same subnet that the iSCSI NIC on the server.
Then in the Targets tab (make sure that you remove all the devices from the Favorites tab - these are the persistent targets), select one of the SPA paths in the Target tab, click connect, check the MPIO and the Reconnect boxes, then click on Advanced button. In the top drop down, select the MS iSCSI Initiator, in the second drop down select the specific Host iSCSI NIC, then in third drop down select the SP port. What this does is "zone" the NIC to the SP port. Click OK, then do the same for the SPB port using the same NIC (remember, you can only use one iSCSI NIC on the host).
Once I get all the paths setup, then I create the Storage Group, add the host, then add the LUNs. Always make sure that when the LUNs are added that you have a HLU 0 (zero) - it should be the first LUN that gets added to the Storage Group.
Having the Host Agent installed also helps with registering the server, setting the failover mode and arraycommpath settings.
Check KB Article 34447 for specific information about connected Windows using iSCSI to CLARiiON arrays
Check KB Article 71615 for all articles about using iSCSI with CLARiiON/VNX arrays.
_https://support.emc.com/kb/34447
_https://support.emc.com/kb/71615
glen
christopher_ime
2K Posts
0
November 8th, 2013 16:00
The order you should do it when you present new LUNs/path from the array is:
1) First rescan within Disk Management (or diskpart)
2) *Then* come back to PowerPath
Or as you have done, just wait as Windows will eventually recognize them on its own. Think of it in layers or more specifically the hierarchy that the LUN must be presented: OS then Application (PowerPath and its drivers)
Also, if you are working with PowerPath exclusively from CLI, you should remember to run the following after you verify PowerPath claims all the expected device paths.
powermt save
As a side-note, for Unix (or Unix-based) systems, you would also want to run: powermt config before the save.
christopher_ime
2K Posts
0
November 8th, 2013 16:00
You are correct, for option "b" using the "Reconnect" action there is no ability to select just one path.
However, for option "c" where this was mentioned, if you go into Engineering mode (again Google for the keystroke combination and the "password"), you will see a new button: "Advanced" when you view the "Hosts" tab of the Storage Group. If you click on that, you'll see individual checkboxes representing the individual paths. There you have the ability to check (or uncheck) at the aforementioned granularity.
wolf8963
46 Posts
0
November 12th, 2013 05:00
Hi Glen,
I was aware of the limitation for the unlicensed powerpath. It does raise a question, however.
Since only 2 paths are supported, should I not even enter the SPA1 and SPB1 paths in the discovery portal and just leave it at SPA0 and SPB0? Or perhaps I should include all in the discovery portal but only log in targets SPA0 and SPB0.
wolf8963
46 Posts
0
November 12th, 2013 05:00
Thanks everyone for your input. I am now seeing drives correctly in disk management, with no offline drives!.
I have discovered that SNMP is seeing the SAN drive types as VM instead of drive. I will start another post on that issue